Oh and I also changed the auto prompt a bit, after florence2 gets the caption and mods it I send it over to ollama advance and w/ a small llm I tell that to "Revise this video prompt to describe the elements of the video, focusing on the motions or actions of the subject in the short clip, as well as the type of camera shot or what the camera is doing. Provide the new prompt in a single paragraph only." So then it's more video focused and not image focused. like I had a photo of a dog and Florence would always say it's mid air and that would always turn into just a dog jumping around to be in mid air...after my ollama add in it would describe the dog as running mostly, also used the pre-prompt to describe it as basic as possible what I wanted "a dog running thru a valley" and it comes out good "A dog running through an open field during golden hour, captured on a sharp 16mm video that highlights its joyful agility and wild energy. The vibrant hues of orange, yellow, and blue add to the dynamic scene, emphasizing both the dog's natural hunting instincts and the natural beauty of the setting."